Oecologia (2015) 178:999–1015 DOI 10.1007/s00442-015-3271-0

PHYSIOLOGICAL ECOLOGY - ORIGINAL RESEARCH

Sex‑ and habitat‑specific movement of an omnivorous semi‑terrestrial controls habitat connectivity and subsidies: a multi‑parameter approach

Lena Hübneṙ · Steven C. Pennings · Martin Zimmer

Received: 4 June 2014 / Accepted: 11 February 2015 / Published online: 19 March 2015 © Springer-Verlag Berlin Heidelberg 2015

Abstract Distinct habitats are often linked through fluxes other high-marsh detrital food sources, and the periwinkle of matter and migration of organisms. In particular, inter- Littoraria was the preferred prey of male (but not female) tidal ecotones are prone to being influenced from both from the forest habitats; both male and female crabs the marine and the terrestrial realms, but whether or not from marsh habitat preferred the fiddler crab Uca over three small-scale migration for feeding, sheltering or reproduc- other prey items. In the field, the major food sources were ing is detectable may depend on the parameter studied. clearly vegetal, but males have a higher trophic position Within the ecotone of an upper saltmarsh in the United than females. In contrast to food preference, isotope data States, we investigated the sex-specific movement of the excluded Uca and Littoraria as major food sources, except semi-terrestrial crab cinereum using an approach for males from the forest, and suggested that Armases con- of determining multiple measures of across-ecotone migra- sumes a mix of C4 and C3 plants along with prey. tion. To this end, we determined food preference, digestive Digestive enzyme activities differed significantly between abilities (enzyme activities), bacterial hindgut communities sexes and habitats and were higher in females and in marsh (genetic fingerprint), and the trophic position of Armases crabs. The bacterial hindgut community differed signifi- and potential food sources (stable isotopes) of males versus cantly between sexes, but habitat effects were greater than females of different sub-habitats, namely high saltmarsh sex effects. By combining multiple measures of feeding and coastal forest. Daily observations showed that Armases ecology, we demonstrate that Armases exhibits sex-specific moved frequently between high-intertidal (saltmarsh) and habitat choice and food preference. By using both coastal terrestrial (forest) habitats. Males were encountered more forest and saltmarsh habitats, but feeding predominantly in often in the forest habitat, whilst gravid females tended to the latter, they possibly act as a key biotic vector of spatial be more abundant in the marsh habitat but moved more subsidies across habitat borders. The degree of contributing frequently. Food preference was driven by both sex and to fluxes of matter, nutrients and energy, however, depends habitat. The needlerush Juncus was preferred over three on their sex, indicating that changes in population structure would likely have profound effects on ecosystem connec- tivity and functioning. Communicated by Liliane Ruess. Keywords Saltmarsh · Coastal forest · Land crab · L. Hübneṙ · M. Zimmer Zoologisches Institut, Christian-Albrechts-Universität zu Kiel, Sexual dimorphism · Spatial subsidy · Habitat 24108 Kiel, Germany connectivity · Motile link organism

S. C. Pennings Department of Biology and Biochemistry, University of Houston, Houston 77204, TX, USA Introduction

* M. Zimmer ( ) Historically, ecologists considered food webs as static rep- Leibniz Center for Tropical Marine Ecology-Mangrove Ecology, Fahrenheitstr. 6, 28359 Bremen, Germany resentations of communities, and habitats as closed systems e-mail: martin.zimmer@zmt‑bremen.de comprising these communities. This approach was useful

1 3 1000 Oecologia (2015) 178:999–1015 for addressing many questions; however, ecosystems and that move between the intertidal and the supratidal may communities are in fact neither closed nor static, and the help sustain nutrient balances in the former. fluxes of matter and migration of organisms across eco- A relatively unstudied aspect of subsidies driven by system boundaries contribute significantly to community motile link organisms is the extent to which these subsidies dynamics, food web dynamics, and ecosystem functioning are sexually dimorphic. Male and female organisms dif- (Polis et al. 1997; Loreau et al. 2002; Knight et al. 2005). fer in many aspects of their behavior and ecology (Breed Spatial subsidies across habitat boundaries (allochthonous et al. 2006; Thaxter et al. 2009; Helfer et al. 2012), and inputs) are ubiquitous and can be mediated by passive so it is likely that they contribute to subsidies in different physical transport or by active biotic movement. For exam- ways, or to different extents, but this has rarely been stud- ple, the deposition of macrophyte wrack of marine origin ied. Furthermore, spatial subsidies are mostly considered ashore in the marine–terrestrial ecotone is driven by wind in one direction, fueling one receptor habitat (sink) from or water movement (Orr et al. 2005). In contrast, many one donor habitat (source), but back-and-forth movements organisms actively move between different habitats that of “motile link organisms” among adjacent habitats may provide feeding grounds versus shelter (Polis et al. 1997; result in bi-directionally subsidizing both habitats, mak- Lewis et al. 2007; Garcia et al. 2011) or change habitats ing a distinction between sink and source difficult or even during their life cycle (e.g., and insects, Knight obsolete. Deciding upon source versus sink habitats may et al. 2005; Gratton et al. 2008; Gratton and Vander Zanden depend upon the parameter under investigation: studying 2009; Hoekman et al. 2011; Vander Zanden and Gratton only one measure of habitat connectivity through motile 2011) and, hence, act as biotic vectors of spatial subsidies links might result in a different picture than using an inte- by translocating material (prey, fecal matter, etc.) between grated approach of quantifying multiple parameters. habitats. These “motile link organisms” (sensu Lundberg Here, we focus on a crab, (hereafter and Moberg 2003) transfer not just nutrients but also eco- Armases), that is common in southeastern USA saltmarshes. logical processes. For example, a motile consumer can Armases is often described as a “saltmarsh” crab (Pennings affect prey densities in multiple habitats, potentially propa- et al. 1998; Buck et al. 2003; Zimmer et al. 2004; Ho and gating interactions from one habitat into the other (Henry Pennings 2008), but it prefers drier and higher areas in the and Jefferies 2009), and food web dynamics in the two hab- intertidal region, and is usually found under wrack and wood itats are linked (McCann 2012). (Seiple 1979) at the terrestrial border of the marsh or in the Tidal saltmarshes are ecologically important (Barbier coastal forest. From there, these crabs range far into either et al. 2011) habitats situated between subtidal marine and habitat (Pennings et al. 1998), but must return to the water to upland habitats. There is an extensive body of research release their eggs during the reproductive period (c.f. Anger examining the physical and biotic mechanisms driving sub- 1995 for terrestrial crabs). Armases is broadly omnivorous sidies between saltmarshes and subtidal systems. Studies of (Buck et al. 2003; Ho and Pennings 2008; Ewers et al. 2012) physical subsidies have focused on “outwelling”, the idea and so may act in translocating both organic matter and con- that a large proportion of the extraordinarily high primary sumer–prey interactions back and forth between the inter- production of saltmarshes is moved by tides into subtidal tidal and supratidal area. Male and female Armases differ habitats, where it fuels heterotrophic coastal metabolism morphologically in that males have larger claws and are bet- (Teal 1962; Valiela and Teal 1974; Peterson and Howarth ter able to eat hard-shelled prey (Buck et al. 2003). Hence, 1987; Krest et al. 2000; Cai 2011) and may affect commu- sex-specific migration patterns or spatiotemporal changes in nity structure (Peterson and Howarth 1987; Currin et al. population structure (sex ratio) may have profound effects 1995; Créach et al. 1997; Fantle et al. 1999; Guest et al. on the ecological role of this motile link organism that con- 2004). Studies of biotic subsidies have focused on juve- nects saltmarshes and coastal forest. The magnitude and nile nekton living in coastal waters next to marshes (Kneib nature of this putative cross-boundary linkage depends on 1997; Jones et al. 2002; Mazumder et al. 2006). In con- how Armases uses adjacent habitats, that is, where and what trast to this extensive body of research, relatively little is Armases crabs feed upon, and how they move between habi- known about biotic subsidies between marshes and adja- tats. Using field and laboratory approaches, we tested the cent upland habitats. It has long been recognized that fish, hypothesis that individual male and female Armases crabs at mammals, alligators and birds move between marshes and Sapelo Island (GA, USA) regularly move between the upper upland habitats to feed and rest (Rosenblatt et al. 2013), saltmarsh and the coastal forest, versus the alternate hypoth- but few studies have quantified the extent of these linkages esis that crabs stay for extended periods in individual habi- (but see Garcia et al. 2011; Brittain et al. 2012; Platt et al. tats, forming stable subpopulations of marsh crabs and for- 2013). Along this line, tidal inundation may remove organic est crabs. For this, we used two complementary approaches matter from the intertidal into coastal waters (c.f. Orr et al. to obtain a comprehensive picture of the potential of salt- 2005, and references therein), but motile link organisms marsh crabs to act as biotic vectors of spatial subsidies:

1 3 Oecologia (2015) 178:999–1015 1001

(1) observations of crab movements in the field, and (2) labo- produce large amounts of litter that accumulates in the high ratory analysis of feeding ecology based on four parameters. marsh (Zimmer et al. 2004). Other plants, like Iva frutescens First, food preferences: preference tests provide informa- Linnaeus and Sarcocornia spp. Linnaeus, are also common tion about potential food preferences, when and if different near (Sarcocornia) or at (Iva) the terrestrial border of Geor- foods are available (which might or might not be the case gia saltmarshes (Pennings et al. 1998) but were not domi- in the field). Second, digestive abilities (enzyme activities): nant in the area where the research was conducted. Hence, the activities of different digestive enzymes inform about Quercus, Borrichia, Juncus and Spartina were used in feed- recently ingested food sources, assuming that most digestive ing experiments, because they were common at the study enzymes are inducible in that their activities correspond with site in different habitats that crabs might move between, and the mixture of food sources inside the digestive tract. Third, because Armases is known to feed on all these plants (Pen- bacterial gut communities (genetic fingerprint): the gut nings et al. 1998; Buck et al. 2003; Zimmer et al. 2004). microbiota are an indicator of sex, habitat, and food sources Four small invertebrates, two gastropods and two crus- (at a range of temporal scales) (Dittmer et al. 2012; Mattila taceans, were common in the study area, and Armases is et al. 2014). And fourth, trophic positions (stable isotopes) also known to feed on these small invertebrates living of crabs found in both habitats: the stable isotope signature either in the terrestrial habitat or in the saltmarsh (Buck of consumers provides a long-term integration of actual food et al. 2003; Ho and Pennings 2008). The periwinkle, Lit- choice in the field. toraria irrorata (Say), is most abundant at lower latitudes, Previous studies have examined species-specific contri- and the coffeebean snail, Melampus bidentatus Say, occurs butions to connecting adjacent habitats. We asked whether along the entire US east coast (Silliman and Zieman 2001). movement and feeding patterns differed among individuals Both snail species were consumed by Armases in previous of the same species depending on sex and habitat. If so, this experiments (Buck et al. 2003), although the amount eaten would reveal the potential for intraspecific variation in con- differed between male and female crabs and depended on tributing to spatially subsidizing neighbor habitats through crab and snail size. Fiddler crabs, Uca spp., are abundant movement and feeding patterns that differ between sexes in lower Georgia saltmarshes (Plumley et al. 1980), with 2 and among subpopulations from adjacent habitats. Moreo- an average density of ca 200 crabs m− (Wolf et al. 1975). ver, if predictable intraspecific variation in subsidies exists, Armases crabs grow rapidly when fed dead fiddler crabs it will allow for predicting changes in spatial connectivity [Uca pugnax (Smith)] (Buck et al. 2003). The common of adjacent habitats upon changes in population structure. amphipod, Orchestia grillus (Bosc), was chosen to repre- sent a high-marsh invertebrate that is known to be readily eaten by both female and male Armases (Buck et al. 2003). Materials and methods Movement in the field Study site and species According to the site of capture, crabs were designated as Field work was performed during May–July 2009 in the either “marsh crabs” or “forest crabs”. We further identi- coastal forest and saltmarsh on western Sapelo Island, GA, fied crabs as males, females or gravid females. An area of USA (31°27′N, 81°15′W). Laboratory work for enzyme meas- 20 m width (parallel to the coastline) and stretching from urements, genetic fingerprinting and preparing stable isotope the marsh into the forest (perpendicular to the coastline) analysis was carried out at the University of Georgia Marine over a length of 25 m (Fig. 1) was divided into 5 m 5 m × Institute (UGAMI) on Sapelo Island and the Zoological Insti- squares and marked with colored flags for orientation. This tute of the Christian-Albrechts-Universtität, Kiel, Germany. area included a 10-m-wide coastal forest area on the inland Saltmarshes around Sapelo Island are typical of the site (forest), a 5-m-wide transition habitat at the terrestrial south-eastern Atlantic coast of the US (Ho and Pennings border (transition) and a 10-m-wide high marsh habitat 2008) and are dominated by the C4 cordgrass, Spartina dominated by Juncus (marsh). These areas will henceforth alterniflora Loisel (Pennings et al. 1998; Zimmer et al. be referred to as forest, transition, and marsh. 2004). The most common high marsh plant is the C3 black To observe the movement of Armases between the habi- needlerush, Juncus roemerianus Scheele (Parsons and De tats, 300 crabs were collected in May 2009 and marked. Of La Cruz 1980; Pennings et al. 1998). The composite shrub these, 150 crabs were collected in the forest and marked Borrichia frutescens (Linnaeus) is common at the terrestrial with yellow nail polish with numbers from 1 to 150. The border (Pennings et al. 1998), and the coastal forest adja- other 150 crabs were collected in the marsh and marked cent to the marsh is dominated by live oak, Quercus vir- with pink nail polish from 1 to 150. All were giniana Miller. Both saltmarsh plants, Spartina and Juncus, scored as males, females or gravid females. Carapace width and plants from the terrestrial habitat, especially Quercus, and length were measured with calipers. We compared the

1 3 1002 Oecologia (2015) 178:999–1015

Fig. 1 The sampling set- ting, ranging from the upper marsh (Juncus) into the forest (Quercus)

size of males and females, and forest crabs and marsh crabs were collected in the area described above and taken to the through pair-wise comparisons (α 0.05) by using Mann– laboratory. The fresh weight of each crab was determined. = Whitney U tests (GraphPad Prism 5). Leaf litter was used because it was preferred over fresh Crabs were released in the center of the marked areas in leaves in previous experiments (data not shown; Buck et al. which they were captured. Marked Armases were observed 2003), but litter was in an early stage of decomposition. once daily for about 1 h for 25 days around slack low Crabs (n 7 for each group) were maintained at room = tide during daytime. In each area of 5 m width and 20 m temperature individually in the laboratory in Petri dishes length, a single investigator searched for marked Armases (14 cm diameter). A small bowl of water (15 psu) was pre- for 15 min while slowly walking the transect. Each num- sent to prevent dehydration of the crabs during the experi- bered crab that was observed was recorded and counted as ment. For each feeding trial, detrital leaves were cut into a “recapture”, although crabs were not physically captured two pieces of approximately same size. One piece was used during daily monitoring. for the treatment, the other piece was used in a crab-free To show the ranges of movement of crabs that were cap- control to monitor autogenic changes in mass (Peterson and tured more than two times during the observation, an indi- Renaud 1989). Fresh weight (FW) of each half was deter- vidual ellipse of the recapture range was created for each mined. Each feeding trial contained four leaves per crab, crab. Ellipses show the centroid of ellipsoid range and one of each plant species. Based on consumption rates as maximum distance moved by the crabs along (horizontally) determined in pilot experiments (not shown), feeding tri- and perpendicular to (vertically) the marsh–forest gradi- als ran for 3 days, and crabs were then released at a sep- ent in the marked area. Statistical analyses were performed arate location to ensure that they were only used once in using the Wilcoxon signed-rank test (SPSS 20) to test for experiments. Leaf remains and leaf pieces from the con- differences in the lengths of the two axes of ellipses (verti- trols were weighed, oven-dried at 60 °C to constant weight, cal vs. horizontal) between forest crabs and marsh crabs, and weighed again. We estimated the initial dry weight of further separated into males and females. the leaves using the fresh:dry weight ratio of the control leaves and calculated mass loss of experimental leaves on Food preference a dry weight basis upon correction for autogenic mass loss of control leaves. Total amount eaten of all food sources Food preference among leaf litter of the four plant species together by one crab was considered 100 % consump- were separately evaluated for forest crabs versus marsh tion. For each crab, we calculated the proportion of feed- crabs and females versus males. Armases and leaf litter ing on each of the four food sources. Statistical analyses of

1 3 Oecologia (2015) 178:999–1015 1003

Table 1 Summary of studied enzymes and their relevance for digestive processes Enzyme group Digestive function Indicative of …

Proteases Proteolytic breakdown (proteolysis) of proteins, poly- and oligo- All trophic groups; more important in carnivores peptides from various (vegetal, animal, microbal) sources Cellulase Hydrolytic breakdown (cellulolysis) of cellulose from vegetal and Herbivores, detritivores microbial sources Amylase Hydrolytic breakdown of starch from vegetal sources Herbivores, detritivores Phenol oxidases Oxidative breakdown of (condensed) tannins, other phenolics, and Herbivores, detritivoresa lignins from vegetal sources Esterases Hydrolytic cleavage of aliphatic and aromatic esters into an acid All trophic group; more important in herbivores and and an alcohol; very diverse group with broad spectrum of, detritivoresb mostly not clearly defined, reactions, e.g., breakdown and detoxification of hydrolyzable tannins a Carnivores also exhibit phenol oxidase activity in the context of immune response, wound-healing and molting () but not inside their digestive tract b Carnivores also exhibit multiple esterase activities, but herbivores and detritivores appear to more directly depend on esterases sensu strictu, whereas carnivores are more in need of lipases feeding preference were performed using the poptools add- for broad categories of compounds; the gut microbial com- on to MS Excel for Resampling and Monte Carlo Analysis munities are potentially quite diverse and so can potentially with 9999 permutations (Frenkel 2004). distinguish among many different types of diets, but the Since Armases prefers invertebrate prey over vegetal link between particular microbial species and particular food sources (Buck et al. 2003), we also investigated food dietary items is often obscure; and the stable isotope signa- preference among potential prey items (dead Uca, Orches- ture provides an indication of diet integrated over a longer tia, Melampus and Littoraria) for forest crabs versus marsh time period, but may have low resolution if multiple food crabs and females versus males. Armases and prey were types have similar isotopic signatures. collected as described above. The fresh weight of each crab For each individual, carapace width and length was was determined, and soft body mass of each prey was esti- measured and fresh weight was determined before crabs mated based on previously determined live mass:soft body were anesthetized individually in ice-water for 5 min. mass ratios. Claws were cut off and stored in 3 mL ethanol (~100 %) Crabs (n 7 for each group) were offered the four until processing. We followed the protocol of Pavasovic = prey items simultaneously for 3 days, and afterwards the et al. (2004), using K/Na-phosphate buffer (pH 6.2) for remaining soft body mass of prey items was estimated. The preparation of the digestive tract. The carapace was opened total amount eaten of all food sources by each crab was and tissue remains around the stomach and above the mid- considered 100 % consumption. For each crab, we calcu- gut gland removed. The mid-gut gland (MG) was dissected lated the proportion of feeding on each of the four food out and washed in buffer (K/Na-phosphate buffer, pH sources. Statistical analyses of food preference were per- 6.2) to minimize the risk of contamination with microbial formed using the poptools add-on to MS Excel for Resa- enzymes which may have been present in the gut (Pavaso- mpling and Monte Carlo Analysis with 9999 permutations vic et al. 2004). After washing, the MG was stored in 1 mL (Frenkel 2004). of the above buffer until use for enzyme measurements. The hindgut was dissected out and stored in 1 mL absolute Digestive capabilities and natural food sources ethanol without removing gut content remains for analyses of microbial diversity. In order to determine actual food sources of male and female crabs from different habitats through an integrated Enzyme activities approach of measuring multiple parameters, 50 crabs (25 forest crabs, 25 marsh crabs) were collected for analysis The activity of gut enzymes provides information on the of enzyme activities, gut microbial communities and sta- nature of major food sources (Table 1). For example, spe- ble isotope signatures. By using multiple indicators of diet, cies that feed on a high-protein diet possess a wide range we obtained a more comprehensive picture than one could of proteolytic enzymes, whereas species that feed on large gain from any one of these. These measures have different amounts of carbohydrates exhibit highly active carbo- strengths and weaknesses. The enzyme activities provide hydrases (Johnston and Freeman 2005). Crustaceans are an indication of recent diet, but are fairly general measures known to possess many of the enzymes required to break

1 3 1004 Oecologia (2015) 178:999–1015 down key nutrients such as proteins and carbohydrates Gut bacterial community composition (Pavasovic et al. 2004). Individual midgut glands were homogenized in 1 mL K/ In addition to species-specific intestinal microbes, the gut Na-phosphate buffer (pH 6.2), and homogenates were cen- microbial community reflects the recent actual diet, and trifuged for 10 min with 10,000g at 4 °C. The supernatant intraspecific differences in gut microbial community com- was diluted by adding 1 mL of K/Na-phosphate buffer (pH position suggest individual differences in food sources 6.2) and stored in aliquots at 20 °C until use for individ- (Dittmer et al. 2012). We used denaturating gradient gel − ual enzyme measurements. electrophoresis (DGGE) to visualize differences in the Total protease activity was measured using azocasein as composition of gut bacterial communities of the crabs col- a substrate following the protocol described by Dìaz-Ten- lected in different habitats and crabs of different sexes. orio et al. (2006) in K/Na-phosphate buffer (pH 6.2). Ali- Hindgut samples (n 6 for each group) stored in etha- = quots of 20 μL of MG extract were added to 480 μL of K/ nol (see above) were centrifuged for 1 min (6000g), and Na phosphate buffer (pH 6.2) and mixed. The reaction was the supernatant ethanol was removed. DNA was extracted started by adding 500 μL of 1 % azocasein solution to the using the Qiagen DNAesy kit following the bench protocol samples and stopped after 1 h by adding 500 μL of 20 % for animal tissues. TCA and maintaining at 0 °C for 10 min. Samples were Following the protocol used by Lachnit et al. (2009) centrifuged at 10,000g at 4 °C for 5 min, and absorbance based on Muyzer et al. (1993), PCR-amplification of bac- was recorded at 366 nm with a spectrophotometer. For the terial 16S rRNA genes of bacterial community DNA pre- blank, 20 % TCA was added before adding MG extracts. sent in gut samples was performed using PuReTaq Ready- Cellulase activity was measured using α-cellulose as To-Go PCR Beads (GE Healthcare) in a total PCR volume substrate (Zimmer 2005). Following the protocol of Zim- of 25 μL, containing 10 pmol of each bacterial primer, mer (2005), 20 mg α-cellulose was added to 200 μL ali- 341F-GC [5′-(CGC CCG CCG CGC GCG GCG GGC quots of MG extract and mixed. The Glucose (HK) Assay GGG GCG GGG GCA CGG GGG GC) CTA CGG GAG Kit (Sigma) was used to quantify the amount of enzymati- GCA GCA G-3′] and 534R (5′-ATT ACC GCG GCT GCT cally released glucose following the manufacturer’s pro- GG-3′). After initial denaturation at 94 °C for 2 min, 15 tocol. 200 μL citrate phosphate buffer containing 0.05 % touchdown cycles (annealing at 65 °C for 40 s, incremental

NaN3 was added to the samples that were then incubated reduction by 1 °C per cycle; elongation at 72 °C for 40 s; on a shaker with 200 rpm for 24 h at room temperature. denaturation at 95 °C for 30 s) were followed by 25 anneal- Samples were centrifuged for 10 min (10,000g, 4 °C) after ing cycles (annealing at 50 °C for 40 s; elongation at 72 °C incubation, and the absorbance of supernatants was meas- for 40 s; denaturation at 94 °C for 30 s) and a final anneal- ured at 340 nm. ing (42 °C for 60 s) and elongation (72 °C for 5 min). The Amylase activity was measured as described for cellu- correct size of the amplified DNA fragments was verified lase but using starch instead of cellulose as substrate. through electrophoresis of 2.5 μL of the PCR reaction vol- Total phenol oxidase activity was measured using pyro- ume in 1.5 % agarose (150 V). catechol as a substrate, following the protocol described by DGGE was performed using double gradient polyacryla- Zimmer (2005). Aliquots of 100 μL of MG extract were mide gels (Lachnit et al. 2009) with a denaturing gradient added to 900 μL of 50 mM pyrocatechol solution and from 40 to 80 % (100 %: 7 M urea and 10 M formamide) mixed. Change in absorbance was recorded at 340 nm and and an acrylamide gradient from 6 to 8 %. Electrophoresis followed at 1-min intervals for 10 min. Relative phenol was run at 60 °C for 13.5 h at 80 V in 0.5 TAE buffer × oxidase activity was determined through linear regression in a CBS Scientific DGGE-2001 system. After electro- analysis (GraphPad Prism 5). phoresis, the gel was stained for 45 min in SYBR Gold® Total esterase activity was measured using 1-naph- (Invitrogen), rinsed for 30 min in 1 TAE buffer, and × thyl acetate as a substrate, modified after He (2003). Ali- photographed under UV light. DGGE gels were analyzed quots of 100 μL of MG extract were added to 900 μL of through the generation of a presence–absence matrix based 50 μM 1NA solution and mixed. Change in absorbance on the individual band pattern. All visible bands in the gel was recorded at 320 nm and followed at 1 min intervals for lane were taken into account and assigned to OTU’s for 10 min. Relative Esterase Activity (REA) was determined further calculation using PRIMER v.6.1.9 (Primer-E). We through linear regression analysis (GraphPad Prism 5). calculated Bray–Curtis values without transformation, and We compared enzyme activities of males, females, for- visualized sample similarities (ANOSIM, two-way crossed est crabs and marsh crabs using two-way ANOVA, with sex with replicates) using cluster analysis and non-metric mul- and habitat as factors (SPSS 20). tidimensional scaling (NMDS).

1 3 Oecologia (2015) 178:999–1015 1005

Isotopic signatures

We used stable isotope analysis to identify carbon sources of Armases, to compare the trophic position of Armases with the trophic positions of potential food sources, and to detect diet differences in crabs from different habitats or of different sexes. A randomly chosen mix of leaf litter from numerous indi- viduals of each species and animals were collected at the study site where fieldwork was conducted. Armases were collected from the forest and marsh; all other species were collected only from their primary habitat. Shells of the snails (n 3 for = each species) were removed, guts were discarded, and muscle Fig. 2 Proportions of males, females and gravid females of the crab, tissue was used for analysis of isotopic signatures. Amphipods Armases cinereum, in captures and recaptures in the forest and marsh: (n 3) were taken as a whole, because Armases fed on whole 100 % captures corresponds to 150 captured and marked crabs in = amphipods during previous experiments. For crabs [Armases each habitat; 100 % recaptures corresponds to the total number of recaptured crabs in each habitat, irrespective of where they had been (n 5 for each group) and Uca (n 3)], claws were cut off = = originally caught. In total, 78 of the forest crabs and 89 of the marsh and muscle tissue was taken out after drying. Samples were crabs were recaptured at least once washed with distilled water and oven-dried at 60 °C to con- stant weight before being ground for 2 min at 60 % speed with a pebble mill (Retsch, Polzin Laborbedarf) and homogenized. Samples were weighed and encapsulated in tin capsules fol- lowing the direction for the preparation of C and N solid sam- ples of the UC Davis Stable Isotope Facility (http://stableiso- topefacility.ucdavis.edu/), where samples were analyzed. We compared isotope values of males and females, and forest crabs and marsh crabs pair-wise (α 0.05) by using Mann– = Whitney U tests (GraphPad Prism 5). In order to estimate the proportional contribution of dif- ferent food sources to the nutrition of crabs, we calculated mixing models using the R package SIAR. For this, we complemented our dataset with average isotope values for benthic diatoms, a potential food source that we had not taken into account in our experiment, using data from Kasai et al. (2004) (for a temperate estuary: Japan), encompassing data provided by Riera et al. (1996) and Kurata et al. (2001) (for saltmarshes: France and Japan, respectively), Doi et al. (2005) (for a temperate estuary: Japan) and Kanaya et al. (2007) (for a temperate brackish lagoon: Japan), as well as Haines (1976a, b) and Haines and Montague (1979) for Fig. 3 Proportional distribution of recaptures of forest crabs (a) and benthic micro-algae (diatoms) at Sapelo Island. marsh crabs (b) in either the habitat of their capture (forest or marsh, respectively), in their respective habitat and the transition zone (forest and transition or marsh and transition, respectively), or over the entire transect Results

Movement in the field The Armases population in the forest was biased towards males (>60 %), whereas the population in the marsh was Captured male crabs were on average 43 % heavier biased towards gravid females (>50 %; Fig. 2). Non-gravid (1.0 0.3 vs. 0.7 0.2 g; ANOVA: p < 0.01) but not larger females were rare in both habitats. The high frequency of ± ± (carapace width: 14 3 vs. 14 2 mm; p 0.45) than gravid females was consistent with previous reports that ± ± = females. Irrespective of their sex, crabs captured in the for- females are gravid from mid-April to November (Seiple est were 25 % heavier (1.0 0.3 g; p < 0.01) and slightly 1979; Seiple and Salmon 1987). ± (7 %) larger (15 1 mm; p < 0.05) than those captured in Most of the forest males were recaptured in the forest ± the marsh (0.8 0.2 g; 14 0.9 mm). (ca 55 %) or transition zone, and only a few (ca 10 %) were ± ± 1 3 1006 Oecologia (2015) 178:999–1015

Fig. 4 Proportions of amounts eaten of different vegetal food sources by male and female forest and marsh Armases in preference tests. 100 % corre- sponds to the total amount eaten by each individual crab. Box plots represent minimum, first quartile, median, third quartile and maximum (n 7). Lower case letters indicate= pair-wise differences; the dotted line indi- cates equal consumption (25 %) of each food source

found in the marsh area (Fig. 3a). Forest females (gravid activity depended on both habitat and sex (protease— and non-gravid) showed less pronounced habitat choice habitat: p < 0.001, sex: p 0.003, interaction: p 0.4; = = and were more likely to be recaptured in the transition zone amylase—habitat: p < 0.001, sex: p 0.001, interac- = or in all habitats including the marsh (~22 %). By contrast, tion: p 0.6; phenol oxidase—habitat: p < 0.001, sex: = marsh crabs commonly were observed in all areas includ- p < 0.001, interaction: p 0.003; esterase—habitat: = ing the forest (males ~42 %, females ~46 %) (Fig. 3b). In p < 0.001, sex: p < 0.001, interaction: p 0.4). Cellulase = sum, marsh crabs of both sexes were quite likely to move activity (Fig. 6b) depended on habitat (p < 0.001) but not into the forest, but forest males were more likely to remain sex (p 0.277). = in the forest than forest crab females “Appendix 1”. Gut bacterial community composition Food preference The bacterial community present in the hindgut differed Male and female crabs from both habitats strongly pre- (p 0.03) between habitats and sex, although similarities = ferred Juncus litter over the other three detrital food sources within habitat (R 0.124) and sex (R 0.209) were low. = = (p < 0.001; Fig. 4). Among the available prey items, males Six of the female crabs analyzed (two of them being for- (both from the forest and the marsh habitat) preferred Lit- est crabs) harbored bacterial communities in their hindguts toraria, whereas marsh crabs (both females and males) pre- that were clearly distinct from those of male crabs (45 % ferred Uca (Fig. 5). Hence, marsh males did not exhibit any similarity), whereas the other females (mostly forest crabs) preference among Littoraria and Uca but preferred both resembled males by >65 % similarity; 9 out of 12 forest over Melampus and Orchestia, and forest females did not crabs clustered together, with only 3 marsh crabs included show any significant food preference. in this cluster of >60 % similarity (Fig. 7). Accordingly, forest crabs and marsh crabs formed distinct groups in Digestive capabilities and natural food sources MDS with an overlap of only four (three males and one female) forest crabs included in the marsh cluster and only Enzyme activities one (male) marsh crab included in the forest cluster (inset in Fig. 7). Overall, crabs of different sexes were more simi- Enzyme activities differed between female and male lar than crabs of different habitats, and sex was a weaker crabs, and between forest crabs and marsh crabs predictor for dissimilarity in the gut microbiota than was (Fig. 6). Except for cellulase activity, all enzyme habitat.

1 3 Oecologia (2015) 178:999–1015 1007

Fig. 5 Proportions of amounts eaten of different animal food sources by male and female forest and marsh Armases in preference tests. 100 % corre- sponds to the total amount eaten by each individual crab. Box plots represent minimum, first quartile, median, third quartile and maximum (n 7). Lower case letters indicate= pair-wise differences; the dotted line indi- cates equal consumption (25 %) of each food source

Isotope signatures but males were characterized by significantly (p 0.043) = higher values (5.9 0.6 ‰) than females (5.2 0.6 ‰). ± ± Carbon isotopic values of Spartina (δ13C 12.97 to The larger variation in carbon signature of marsh males = − 13.36 ‰) were different from those of transition and than of other crabs suggests a greater variety of food − upland plant litter from Quercus, Borrichia and Jun- sources among male Armases that live in the high marsh. cus (δ13C 25.70 to 31.06 ‰) (Fig. 8). Carbon iso- According to the diet mixing model used herein, for- = − − tope values of potential gastropod prey were clearly est males ate approximately the same amount (c. 5–15 %, separated from those of the potential prey. each) of all 9 food sources considered (“Appendix 2”). Melampus (δ13C 14.91 to 15.25 ‰) and Littoraria All other crabs ingested significantly more Quercus, and = − − (δ13C 14.35 to 15.57 ‰) both had carbon isotopic to a lesser degree Juncus, Orchestia and Melampus (for- = − − values similar to Spartina, whereas values of Orchestia est females) or Borricchia (marsh crabs) than of Littoraria, (δ13C 18.82 to 20.08 ‰) and Uca (δ13C 20.08 Uca or benthic diatoms. = − − = − to 22.96 ‰) had values similar to those of Armases − (δ13C 16.55 to 21.64 ‰). The mean carbon iso- = − − tope value of forest crabs (δ13C 20 1 ‰) and the Discussion = − ± mean value of marsh crabs ( 20.9 0.5 ‰) did not differ − ± (p 0.1) and fell between terrestrial and saltmarsh plants. Movement in the field = Nitrogen isotope values of plants ranged from δ15N 0.77 ‰ in Quercus to δ15N 3.04 ‰ in Spar- Based on our data on spatial distribution and migration of = − = tina. Spartina receives nutrients from coastal water bodies male and female crabs within the marsh–forest ecotone, and is therefore enriched in heavy nitrogen relative to ter- we propose that habitat choice (marsh vs. forest) is sex- restrial plants (Currin et al. 1995). Nitrogen isotope values specific. Males prefer the coastal forest habitat, whereas of Orchestia (1.10–1.89 ‰) and Melampus (1.41–1.84 ‰) gravid females prefer the marsh habitat. Captures of non- were lower than those of Armases, whereas those of Lit- gravid females were low, due to the synchronized breeding toraria (5.92–6.35 ‰) and Uca (5.68–6.20 ‰) were similar period, but they seem to exhibit only weak selectivity for to those of Armases (Fig. 8). Mean nitrogen isotopic values either habitat. Crabs found in the marsh, i.e. mostly gravid did not differ significantly (p 0.3) between forest crabs females, exhibit a large activity range, including both the = (δ15N 5.7 0.9 ‰) and marsh crabs (5.5 0.5 ‰), marsh and the coastal forest. Male crabs, by contrast, and = ± ± 1 3 1008 Oecologia (2015) 178:999–1015

30 0.015 (B) cellulase (A) protease -1 -1

h) 20

n) 0.010 c mi

c (g (g gl

U b 0.005 ab 10 mg RA a

0.000 0 males femalesmalesfemales males females males females forest marsh forest marsh

250 0.03 (C) amylase (D) phenol oxidase 200 -1 -1 b h) n) 0.02 150 mi

(g b

c (g

gl 100 a

U 0.01 a mg a RA 50 a a a 0 0.00 males females malesfemales males females males females forestmarsh forest marsh

0.008 (E) esterase

-1 0.006

n) b mi

(g 0.004 ab U RA 0.002 a a 0.000 males femalesmales females forestmarsh

Fig. 6 Digestive enzyme activities in midgut gland extracts of male tile, median, third quartile and maximum (n 12). Lower case letters and female forest and marsh crabs. a protease, b cellulase, c amylase, indicate pair-wise differences = d phenol oxidase, e esterase. Box plots represent minimum, first quar- particularly those caught in the forest, exhibit a much 1988). However, both male and female crabs move into the more restricted activity range that largely excludes the marsh habitat and, according to our findings on digestive marsh (“Appendix 1”). Nonetheless, our data indicate that capabilities and nutrition, do so for feeding. It thus remains Armases moves regularly between the upper marsh and the an essentially unanswered question why male crabs remain forest and does not establish highly specialized subpopula- in the drier and cooler coastal forest with (detrital) food tions in the forest versus in the marsh. Hence, Armases has sources of relatively low nutritive value; however, forest the potential to act as a vector for organic matter between males seem to have access to a more diverse diet than their the saltmarsh and adjacent coastal forests, but both the type conspecifics that spend more time in the marsh, and this of matter and its spatial fate will depend on the crab’s sex. may have advantages that were not captured in this study. The choice of the marsh habitat by gravid females could be related to the vicinity to water where they release their Integrated feeding ecology larvae after incubation (Lee 1998). Moreover, exposed eggs are likely vulnerable to desiccation in the drier forest hab- According to food preference, enzyme activities and stable itat. Additionally, gravid female crabs as well as fish and isotope signatures, diet choice by Armases reflects habitat shellfish may use the marsh, because dense plant cover in choice which, in turn, is sex-specific. Male crabs in the for- the marsh protects them from predators (McIvor and Odum est equally feed on detritus and invertebrate prey that they

1 3 Oecologia (2015) 178:999–1015 1009

food sources of Armases individuals from the upper marsh and females from the forest at this study site. This distinct preference may be due to the specific habitat that we used to collect and study Armases. This site had large stands of Juncus but relatively little Spartina (that is more common at the low marsh), with Armases frequently being encountered underneath mats of Juncus litter (authors’ personal observa- tions). Quercus litter, on the other hand, although common, appears to be a detrital food source that Armases prefers not to eat in food preference tests (c.f. Zimmer et al. 2004; Ewers et al. 2012). Zimmer et al. (2004), by contrast, found that Armases ate similar amounts of Juncus, Spartina, and Quercus, and much more of Borrichia than reported here. Similarly, the present stable isotope data indicate, in contrast to what we would have concluded from food preference tests, that Quercus (along with Juncus and Borricchia) makes up Fig. 7 Cluster analysis (Euclidean distance) and non-metric MDS a major proportion of the diet of all crabs but forest males. (inset Bray-Curtis similarity) of gut bacterial community composition in male (filled symbols) and female (open symbols) marsh crabs (tri- These previous food preference experiments (Zimmer et al. angles) versus forest crabs (circles) 2004), however, were performed over 30 days, and were no- choice rather than multiple-choice feeding trials, in which animals offered low-quality diets may have been making the best of a bad situation by compensatory consumption. Addi- tionally, Armases had poor survival on Borrichia in these experiments, possibly due to toxic compounds released from Borrichia litter during decomposition. Such effects of Borri- chia, however, are unlikely to occur in the field (Zimmer et al. 2004), where leached compounds would be rapidly diluted. Both Littoraria and Uca contributed insignificantly to the nutrition of crabs, except for forest males, partly corroborat- ing preference tests with (dead) animal prey. While the ability to capture and handle prey will drive food preference in the field, we were not interested in prey capture abilities in our preference test in the laboratory. Hence, we offered dead prey items, possibly explaining why marsh crabs preferred Uca that was only rarely eaten in the field. Among the animal prey items offered, Littoraria were Fig. 8 Stable isotope (15N and 13C) signatures of male (filled sym- bols) and female (open symbols) Armases from marsh (triangles) preferred by males. Female Armases have smaller claws versus forest (circles) habitat and selected potential animal (dashed than males, and therefore have difficulty crushing the hard elipsoid) and vegetal (dotted elipsoid) food sources shell of Littoraria (Buck et al. 2003). Males residing in the forest, on the other hand, would be expected not to be used to capturing and eating Uca (as corroborated by food prefer- can handle with their larger claws. Females, particularly ences), since they hardly ever move into the area of the low gravid ones, in the marsh feed more actively, both on detri- marsh where Uca is common. Both females and males that tus and those more protein-rich animals that are easy to do move into the marsh are experienced in preying upon handle (amphipods). Uca by first cutting off the large claw upon attack and then Previous feeding experiments demonstrated that Armases killing the undefended fiddler crab (personal observation). feeds, and is able to grow on, a wide variety of food sources However, according to our stable isotope data, Uca was not (Pennings et al. 1998; Buck et al. 2003; Zimmer et al. 2004). among the preferred food sources of marsh crabs or forest For other grapsid saltmarsh crabs (Paragrapsus laevis and females, whereas they contributed to the nutrition of forest Helogarapsis haswellianus), microphytobenthic food sources males as much as all other food taken into account herein. have been shown to play a primary dietary role (Alderson Changes in an individual’s diet seem to mediate individ- et al. 2013). Our present results, however, clearly indicate that ual enzyme activities when proportions of components (pro- Juncus detritus was both the preferred and one of the major tein, cellulose, etc.) in the diet change. Such a link between

1 3 1010 Oecologia (2015) 178:999–1015 dietary composition and the activity of digestive enzymes males and females, with females trading-off the availability was shown in many previous studies (e.g., Johnston and of food sources higher than the risk of being preyed upon Freeman 2005) and has been hypothesized for other crusta- by saltmarsh predators, but males using the forest as shelter ceans, such as the mud crab, Scylla serrate (Pavasovic et al. from predation as pay-off for reduced food availability in the 2004). In contrast to stable isotope signatures that integrate forest. over a longer time, this link is short term and indicates most Pavasovic et al. (2004) demonstrated that cellulase activi- recently consumed food types. Male and female Armases ties in mud crabs, Scylla serrata, that fed on starch or cel- differed from each other in that females and crabs captured lulose were up to four times higher than in crabs feeding on in the marsh exhibited higher enzyme activities than males other diets, reflecting a general ability to modulate individ- and crabs captured in the forest, respectively. Protease activ- ual enzyme activities in response to changing proportions of ity is expected to be high in individuals consuming large dietary materials. Plant material of high cellulose content is amounts of protein-rich diet and was expected to be found available in both habitats, and Armases feeds on fresh leaves in Armases, because these crabs are considered omnivorous, and leaf litter of various plants common either in the coastal feeding on both animal and vegetal tissue (but see below). forest, at the terrestrial border between marsh and forest, The higher protease activity in crabs captured in the marsh or in the marsh (Buck et al. 2003; Zimmer et al. 2004; Ho corroborates reports of Armases feeding on invertebrate her- and Pennings 2008). The major vegetal food sources seem bivores of saltmarsh plants (Ho and Pennings 2008; Marczak to be Juncus, Borricchia and Quercus that can be found in et al. 2011) or on worms, insect larvae and other invertebrates the marsh and forest. Like cellulases, amylases are required that are present in the mud of saltmarshes (Buck et al. 2003, for the digestion of plant material (starch), but, in contrast to Zimmer et al. 2004), all being rich in protein. This assump- cellulose, they also break down glycogen in animal tissue. tion supports our hypothesis that crabs move into the marsh Similar to proteases and in contrast to cellulases, amylase for feeding because these protein-rich food sources are abun- activity was highest in crabs captured in the marsh. Thus, dant and easier to find in muddy marsh sediments than in amylase in Armases seems to be induced upon feeding on forest soils. High protease activity in crabs sampled in the invertebrate prey, such as Orchestia, insects, snails or nema- marsh suggests that they had been feeding on marsh inverte- todes, for cleaving α-1,4 linkages in glycogen, as has been brates shortly before they were captured, but among the ani- proposed for the mud crab, Scylla serrata (Pavasovic et al. mal prey tested here, only Orchestia significantly contributed 2004) and slipper lobsters (Johnston and Yellowlees 1998). to the nutrition of marsh crabs in the long term. Along the Differences between males and females may, thus, be due to same line, (gravid) females that tended to be more abundant sex-specific feeding strategies, with females (in the marsh) in the marsh would make use of a higher proportion of inver- consuming more of these food sources than males. tebrate prey items, and, thus, exhibit higher protease activ- Feeding on vegetal food sources is often constrained by ity. Buck et al. (2003) showed that sexual dimorphism in high tannin contents. One way for consumers to deal with claw size influences predatory ability in Armases. Although tannins in vegetal tissue is to feed on leached leaf litter male crabs are able to feed on better-defended prey (here Lit- (Giddins et al. 1986; Schofield et al. 1998; Zimmer 2002). toraria), females may feed on other prey (insect larvae, etc.) Armases showed a clear preference for leaf litter over fresh that are also rich in protein. However, δ15N-values of male leaves (author’s own observation; Buck et al. 2003), but versus female crabs suggest that females make their living preferences among different decomposition stages were not based less on predation than males. Corroborating this, John- tested, and Nordhaus (2004) reported that most herbivo- ston and Freeman (2005), investigating dietary preference rous crabs have no preference for consuming aged leaves and digestive enzyme activities in six crab species, showed (with phenolic compounds leached off) over fresh ones, that the highest protease activity occurred in omnivorous and therefore use other mechanisms to deal with phenolic crabs feeding on large amounts of microalgae in contrast to compounds. Esterases are important digestive enzymes for the expected highest protease activity in carnivorous crabs. detritivorous, herbivorous or omnivorous species that have However, benthic diatoms did not significantly contribute to deal with these compounds. Esterase activities did not to the nutrition of female crabs. Kyomo (1992) investigated differ among crabs captured in the marsh versus the forest differences in feeding habitats and feeding modes between but were higher in females than in males, supporting our males and females of the crab intermedia de Haan conclusion that females were generally more engaged in (Crustacea; ) and demonstrated two-times higher feeding than males. The relatively lower contribution of feeding activity (based on claw movement) in females than animal prey to the diet of (female) crabs captured in the in males. Possibly, reproductively active females (as here) marsh would then correspond with the higher phenol oxi- are in need of higher input of nutrients. Such different feed- dase activities in these crabs that encounter large amounts ing abilities and strategies may contribute to the spatial seg- of vegetal detritus of both forest and marsh that accumu- regation between forest and marsh that we observed among lates in the upper marsh (Zimmer et al. 2002, 2004).

1 3 Oecologia (2015) 178:999–1015 1011

Although individual crabs that had been captured in the Armases can frequently be observed sieving through the same region of our study area differed remarkably in the com- saltmarsh sediment (authors’ personal observations), pos- position of their gut bacterial communities, differences among sibly foraging for small detrital particles (POM) or small crabs that had been captured in different regions (forest vs. invertebrate prey (snails, nematodes), but—according to marsh) were even bigger. Taking into account that detrital our present findings—probably less so for microalgae, pos- matter of different plants differ in their bacterial communi- sibly because of their small size. They exert indirect posi- ties (Myers et al. 2001) and different vegetal food sources of tive effects on saltmarsh plants by feeding on their inver- saltmarsh detritivores shape their gut bacterial communities tebrate herbivores (Ho and Pennings 2008; Marczak et al. (Dittmer et al. 2012), this finding suggests that food sources 2011). On the other hand, Armases is known to feed on ter- of crabs captured in the marsh differed from those of crabs restrial plants (Ho and Pennings 2008), and to contribute to captured in the forest. Crabs that were feeding in the marsh the mass loss of saltmarsh litter of terrestrial origin (Zim- harbored more diverse bacterial communities than crabs in the mer et al. 2004; but see Ewers et al. 2012; Treplin et al. forest. Along the same line, females differed from males in 2013). Earlier studies that reported δ13C values from 12.2 − their gut bacteria, with males being more similar to each other to 19.1, depending on the study region, suggest that − than females, suggesting either a higher proportion of sex-spe- crabs differ substantially in what they feed on among sites cific indigenous bacteria in less actively feeding males than in (Haines 1976a; Haines and Montague 1979). In the present heavily feeding females, or a homogenizing effect of staying study, admittedly, numerous additional or alternative food in the forest versus migrating back and forth and feeding on sources may have been used by Armases for which we did more variable food sources as females seem to do. not obtain isotope data. The preference of males for Littoraria is surprising, Whereas male crabs tended to exhibit higher δ15N- taking into account previous findings on preferences for values than females, suggesting a greater contribution of Orchestia (Buck et al. 2003) and Melampus (c.f. Ewers et al. animal versus vegetal matter to their diet, it was impossible 2012). However, stable nitrogen isotope signatures indicate to distinguish between crabs captured in the marsh versus that neither Uca nor Littoraria contribute significantly to the the forest based on δ15N-values. Taking food preferences nutrition of Armases in the field (except for forest males). in the laboratory also into account, we conclude that crabs, While both are preferred nitrogen sources when accessible both males and females, move into the marsh to feed on (as in our laboratory assay), the hard shell of Littoraria pre- Juncus, Borrichia and Quercus, as well as on Orchestia and sents a formidable defense in nature, and Uca may escape by Melampus (Buck et al. 2003) and probably also other inver- fleeing or hiding in their burrows (which was prevented in tebrates. The better suitability of male claws for predatory the laboratory by our experimental design). Hence, although feeding is reflected by the higher δ15N-values of males. Armases is capable of preying upon these prey items, their Hence, the appearance of Armases as an omnivore seems actual contribution to the nutrition of Armases is insignifi- to reflect the females’ preference for vegetal food sources, cant, because Armases choose to feed on easier prey. with males occasionally consuming animal tissue. The carbon signature of Armases suggests that it actually feeds on a mixture of C4-based (Spartina) and C3-based (Jun- cus, Borrichia, Quercus) vegetal sources. Taking into account Conclusions that they may additionally prey on Orchestia (c.f. Buck et al. 2003) and Melampus, these would then make a contribution We conclude that Armases has the potential to act as a of Spartina to their diet less likely, based on the predictions of biotic vector of spatial subsidies. These motile link organ- the mixing model applied. One problem in estimating carbon isms frequently migrate between the marsh and the adja- sources in saltmarsh systems that is discussed in different stud- cent forest, but forest crabs seem to be more stationary, at ies (Haines 1976a, b; Currin et al. 1995; Créach et al. 1997; least during summer. According to this and to our findings Fantle et al. 1999) is that benthic microalgae, and particularly on crabs mostly feeding in the marsh, the forest would be a diatoms, have carbon isotopic values (~ 15 to 20 ‰) that net recipient fed from the marsh as a net donor for spatial − − are similar to values of a mixture of terrestrial and saltmarsh subsidy vectored by crabs. The resulting net flux of matter, plants (Deegan and Garritt 1997) and have been found to play nutrients and energy from the upper saltmarsh into coastal a key role in supporting saltmarsh food webs (Fantle et al. forests may depend on seasonal changes in migratory and/ 1999). These authors concluded that the majority of consum- or feeding behavior, though. Hence, detailed studies on ers appear to have a mixed diet of Spartina and microalgae, whether this subsidy actually occurs are warranted. and Sullivan and Moncreiff (1990) reported microalgae to be Females show higher migration activity than males even more important. Our present data, however, overall sug- and appear to be more active in feeding, and the more gest an insignificant role of both Spartina and benthic diatoms stationary males may form locally stable sub-groups of in the diet of saltmarsh crabs of Sapelo Island. crabs. Food sources encompass both vegetal and animal

1 3 1012 Oecologia (2015) 178:999–1015 matter, being reflected by enzyme activities and stable spatial subsidies through motile link organisms should be isotope signatures. However, the potentially most pre- taken into account in studies of habitat connectivity. ferred prey items, Littoraria and Uca, do not contribute to the natural diet of Armases in the marsh. The most Author contribution statement M.Z. and L.H. con- abundant type of detritus is also one of those preferred ceived and designed the study. L.H. and S.C.P. performed by Armases. Hence, Juncus (and to a lesser degree Bor- the study. L.H. and M.Z. analyzed the data. M.Z. and S.C.P. richia) strongly contributes to the vector-based inter- wrote the manuscript. tidal spatial subsidy to the terrestrial realm. By contrast, spatial subsidies from coastal forests into the marsh are based on Quercus but will play a minor role, due to forest Acknowledgments We thank Stuart Linton (Deakin University, crabs migrating less frequently into the marsh. Intraspe- Australia) for valuable discussion on digestive enzymes; Melissa Both, Mario Muscarella and Daniel Saucedo (UGAMI), and Steph- cific variation with respect to feeding seems to be sex- anie Stratil and Tim Lachnit (GEOMAR, Kiel, Germany) for support specific rather than habitat-specific, but habitat choice and assistance in the field and the lab. We thank NSF (OCE06-20959) also largely depends on the sex (and reproductive status). for funding. This work is contribution #1038 of the University of As can be deduced from most studies that showed sex- Georgia Marine Institute, and a contribution of the Georgia Coastal Ecosystems Long-Term Ecological Research program. specific behavior in migrating species (e.g., Breed et al. 2006; Thaxter et al. 2009; Helfer et al. 2012), male and female Armases will contribute to subsidies in different Appendix 1: Spatial distribution of individual male ways, or to different extents. Hence, any change in popu- and female crabs lation structure or sex ratio will have profound effects on ecosystem connectivity and functioning. This aspect of Spatial distribution of individual male and female crabs motile link ecology has essentially been neglected in previ- according to recapture-based migration patterns within the ous studies but should be taken into account when study- marsh–forest ecotone. Elipsoids reflect the individual range ing biotic vectors of spatial subsidies. Further, as is obvi- of recapture spots along the two axes parallel versus per- ous from our partly contrasting results, several measures of pendicular to the coast line.

1 3 Oecologia (2015) 178:999–1015 1013

Appendix 2: Estimated proportional contribution M, Melampus; L, Littoraria; O, Orchestia; U, Uca; D, of different food sources benthic diatoms) to the nutrition of male and female Armases from marsh versus forest habitat. Box plots Estimated proportional contribution of different food represent minimum, 5, 25, 75 and 95 % percentiles, and sources (J, Juncus; B, Borrichia; S, Spartina; Q, Quercus; maximum.

References food types, body size and habitat complexity. J Exp Mar Biol Ecol 292:103–116 Alderson B, Mazumder D, Saintilan N, Zimmerman K, Mulry P Cai W-J (2011) Estuarine and coastal ocean carbon paradox: CO2 (2013) Application of isotope mixing models to discriminate sinks or sites of terrestrial carbon incineration? Annu Rev Mar dietary sources over small-scale patches in saltmarsh. Mar Ecol Sci 3:123–145 Progr Ser 487:113–122 Créach V, Schricke MT, Bertru G, Mariotti A (1997) Stable isotopes Anger K (1995) The conquest of freshwater and land by marine crabs: and gut analyses to determine feeding relationships in saltmarsh adaptations in life-history patterns and larval bioenergetics. J Exp macroconsumers. Estuar Coast Shelf Sci 44(5):599–611 Mar Biol Ecol 193:119–145 Currin CA, Newell SY, Paerl HW (1995) The role of standing dead Barbier EB, Hacker SD, Kennedy C, Koch E, Stier AC, Silliman BR Spartina alterniflora and benthic microalgae in saltmarsh food (2011) The value of estuarine and coastal ecosystem services. webs—considerations based on multiple stable isotope analysis. Ecol Monogr 81:169–193 Mar Ecol Prog Ser 121:99–116 Breed GA, Jonsen ID, Myers RA, Bowen WD, Leonard ML (2006) Deegan LA, Garritt RH (1997) Evidence for spatial variability in Sex-specific, seasonal foraging tactics of adult grey seals (Hali- estuarine food webs. Mar Ecol Prog Ser 147:31–47 choerus grypus) revealed by state-space analysis. Ecology Dìaz-Tenorio LM, García-Carreno FL, Navarrete del Toro MA (2006) 90:3209–3221 Characterization and comparison of digestive proteinases of the Brittain RA, Schimmelmann A, Parkhurst DF, Craft CB (2012) Habi- Cortez swimming crab, Callinectes bellicosus, and the arched tat use by coastal birds inferred from stable carbon and nitrogen swimming crab, Callinectes arcuatus. Invertebr Biol 125:125–135 isotopes. Estuar Coasts 35:633–645 Dittmer J, Lesobre J, Raimond R, Zimmer M, Bouchon D (2012) Buck TL, Breed GA, Pennings SC, Chase ME, Zimmer M, Carefoot Influence of changing plant food sources on the gut microbiota of TH (2003) Diet choice in an omnivorous saltmarsh crab: different saltmarsh detritivores. Microb Ecol 64:814–825

1 3 1014 Oecologia (2015) 178:999–1015

Doi H, Matsumasa M, Toya T, Satoh N, Mizota C, Maki Y, Kikuchi E Kasai A, Horie H, Sakamoto W (2004) Selection of food sources by (2005) Spatial shifts in food sources for macrozoobenthos in an Ruditapes philippinarum and Mactra veneriformis (Bivalva: Mol- estuarine ecosystem: carbon and nitrogen stable isotope analyses. lusca) determined from stable isotope analysis. Fish Sci 70:11–20 Estuar Coast Shelf Sci 64:316–322 Kneib RT (1997) The role of tidal marshes in the ecology of estuarine Ewers C, Beiersdorf A, Wieski K, Pennings SC, Zimmer M (2012) nekton. Oceanogr Mar Biol Annu Rev 35:163–220 Predator/prey-interactions promote decomposition of low-quality Knight TM, McCoy MW, Chase JM, McCoy KA, Holt RD detritus. Wetlands 32:931–938 (2005) Trophic cascades across ecosystems. Nature Fantle MS, Dittel AI, Schwalm SM, Epifanio CE, Fogel ML (1999) 437(7060):880–883 A food web analysis of the juvenile blue crab, Callinectes sapi- Krest JM, Moore WS, Gardner LR, Morris JT (2000) Marsh nutrient dus, using stable isotopes in whole animals and individual amino export supplied by groundwater discharge: evidence from radium acids. Oecologia 120:416–426 measurements. Glob Biogeochem Cycles 14:167–176 Frenkel D (2004) Introduction to Monte Carlo Methods. In: Attig N, Kurata K, Minami H, Kikuchi E (2001) Stable isotope analy- Binder K, Grubmüller H, Kremer K (eds) Computational soft sis of food sources for salt marsh snails. Mar Ecol Prog Ser matter: from synthetic polymers to proteins. John von Neumann 223:167–177 Institute for Computing, Jülich, pp 23–59 Kyomo J (1992) Variations in the feeding habits of males and Garcia EA, Bertness MD, Alberti J, Silliman BR (2011) Crab regula- females of the crab Sesarma intermedia. Mar Ecol Prog Ser tion of cross-ecosystem resource transfer by marine foraging fire 83:151–155 ants. Oecologia 166:1111–1119 Lachnit T, Bümel M, Imhoff JF, Wahl M (2009) Specific epibacterial Giddins RL, Lucas JS, Neilson MJ, Richards GN (1986) Feeding communities on macroalgae: phylogeny matters more than habi- ecology of the mangrove crab (Crustacea: tat. Aquat Biol 5:181–186 : ). Mar Ecol Prog Ser 33:147–155 Lee SY (1998) Ecological role of grapsid crabs in mangrove ecosys- Gratton CJ, Vander Zanden MJ (2009) Flux of aquatic insect produc- tems: a review. Mar Freshw Res 49:335–343 tivity to land: comparison of lentic and lotic ecosystems. Ecology Lewis TL, Mews M, Jelinski DE, Zimmer M (2007) Detrital subsidy 90:2689–2699 to the supratidal zone provides feeding habitat for intertidal crabs. Gratton CJ, Donaldson J, Vander Zanden MJ (2008) Ecosystem link- Estuar Coast 30:451–458 ages between lakes and the surrounding terrestrial landscape in Loreau M, Naeem S, Inchausti S (2002) Biodiversity and ecosystem northeast Iceland. Ecosystems 11:764–774 functioning: synthesis and perspectives. Oxford University Press, Guest MA, Connolly RM, Loneragan NR (2004) Carbon movement Oxford and assimilation by invertebrates in estuarine habitat at a scale of Lundberg J, Moberg F (2003) Mobile link organisms and ecosystem metres. Mar Ecol Prog Ser 278:27–34 functioning: implications for ecosystem resilience and manage- Haines EB (1976a) Relation between the stable carbon isotope com- ment. Ecosystems 6:87–98 position of fiddler crabs, plants and soils in a saltmarsh. Limnol Marczak LB, Ho C-K, Wieski K, Vu H, Denno RF, Pennings SC Oceanogr 21:880–883 (2011) Latitudinal variation in top-down and bottom-up control Haines EB (1976b) Stable carbon isotope ratios in the biota, soils and of a saltmarsh food web. Ecology 92:276–281 tidal water of a Georgia saltmarsh. Estuar Coast Mar Sci 4:609–616 Mattila JM, Zimmer M, Vesakoski O, Jormalainen V (2014) Habitat- Haines EB, Montague CL (1979) Food sources of estuarine inverte- specific gut microbiota of the marine herbivore Idoteabalthica brates analyzed using 13C/12C ratios. Ecology 60:48–56 (Isopoda). J Exp Mar Biol Ecol 455:22–28 He X (2003) A continuous spectrophotometric assay for the determi- Mazumder D, Saintilan N, Williams RJ (2006) Trophic relationships nation of diamondback moth esterase activity. Arch Insect Bio- between itinerant fish and crab larvae in a temperate Australian chem Physiol 54:68–76 saltmarsh. Mar Freshw Res 57:193–199 Helfer V, Broquet T, Fumagalli L (2012) Sex-specific estimates of McCann KS (2012) Food webs. Princeton University Press, Princeton dispersal show female philopatry and male dispersal in a promis- McIvor CC, Odum WE (1988) Food, predation risk, and microhabitat cuous amphibian, the alpine salamander (Salamandra atra). Mol selection in a marsh fish assemblage. Ecology 69:1341–1351 Ecol 21:4706–4720 Muyzer G, de Waal EC, Uitterlinden AG (1993) Profiling of complex Henry HAL, Jefferies RL (2009) Opportunistic herbivores, migratory microbial populations by denaturing gradient gel electrophoresis connectivity, and catastrophic shifts in Arctic coastal systems. In: analysis of polymerase chain reaction-amplified genes coding for Silliman BR, Grosholz ED, Bertness MD (eds) Human impacts 16S rRNA. Appl Environ Microbiol 59:695–700 on saltmarshes: a global perspective. University of California Myers RT, Zak DR, White DC, Peacock A (2001) Landscape-level Press, Berkeley, pp 85–102 patterns of microbial community composition and substrate use Ho C-K, Pennings SC (2008) Consequences of omnivory for trophic in upland forest ecosystems. Soil Sci Soc Am J 65:359–367 interactions on a saltmarsh shrub. Ecology 89(6):1714–1722 Nordhaus I (2004) Feeding ecology of the semi-terrestrial crab Ucides Hoekman D, Dreyer J, Jackson R, Townsend P, Gratton CJ (2011) cordatus (Decapoda: Brachyura) in a mangrove forest in northern Lake to land subsidies: experimental addition of aquatic insects Brazil. PhD thesis. University of Bremen increases terrestrial densities. Ecology 92:2063–2072 Orr M, Zimmer M, Mews M, Jelinski DE (2005) Wrack deposition on Johnston DJ, Freeman J (2005) Dietary preference and digestive different beach types: spatial and temporal variation in the pattern enzyme activities as indicators of trophic resource utilization by of subsidy. Ecology 86:1496–1507 six species of crab. Biol Bull 208:36–46 Parsons KA, De la Cruz AA (1980) Energy flow and grazing behavior Johnston DJ, Yellowlees D (1998) Relationship between dietary prefer- of conocephaline grasshoppers in a Juncus roemerianus marsh. ence and digestive enzyme complement of the slipper lobster, The- Ecology 6:1045–1050 nus orientalis (Decapoda: Scyllaridae). J Crustac Biol 18:656–665 Pavasovic M, Richardson NA, Anderson AJ, Mann D, Mather Jones RF, Baltz DM, Allen RL (2002) Patterns of resource use by PB (2004) Effect of pH, temperature and diet on digestive fishes and macroinvertebrates in Barataria Bay, Louisiana. Mar enzyme profiles in the mud crab, Scylla serrata. Aquaculture Ecol Prog Ser 237:271–289 242:641–654 Kanaya G, Takagi S, Nobata E, Kikuchi E (2007) Spatial dietary shift Pennings SC, Carefoot TH, Siska EL, Chase ME, Page TA (1998) of macrozoobenthos in a brackish lagoon revealed by carbon and Feeding preferences of a generalist salt-marsh crab: relative nitrogen stable isotope ratios. Mar Ecol Prog Ser 345:117–127 importance of multiple plant traits. Ecology 79:1968–1979

1 3 Oecologia (2015) 178:999–1015 1015

Peterson BJ, Howarth RW (1987) Sulfur, carbon and nitrogen iso- Sullivan MJ, Moncreiff S (1990) Edaphic algae are important compo- topes used to trace organic matter flow in the salt-marsh estuaries nent of saltmarsh food-webs: evidence from multiple stable iso- of Sapelo Island, Georgia. Limnol Oceanogr 32:1195–1213 tope analyses. Mar Ecol Prog Ser 62:149–159 Peterson CH, Renaud PE (1989) Analysis of feeding preference Teal JM (1962) Energy flow in the saltmarsh ecosystem of Georgia. experiments. Oecologia 80:82–86 Ecology 43:614–624 Platt SG, Elsey RM, Liu H, Rainwater TR, Nifong JC, Rosenblatt Thaxter CB, Daunt F, Hamer KC, Watanuki Y, Harris MP, Grémillet AE, Heithaus M, Mazzotti FJ (2013) Frugivory and seed disper- D, Peters G, Wanless S (2009) Sex-specific food provisioning sal by crocodilians: an overlooked form of saurochory? J Zool in a monomorphic seabird, the common guillemot Uria aalge: 291:87–99 nest defence, foraging efficiency or parental effort? J Avian Biol Plumley FG, Davis DE, McEnerney JT, Everest JW (1980) Effects of 40:75–84 a photosynthesis inhibitor, atrazine, on the saltmarsh fiddler crab, Treplin M, Pennings SC, Zimmer M (2013) Decomposition in a US Uca pugnax. Estuaries 3:217–223 saltmarsh is driven by dominant species not species complemen- Polis GA, Anderson WB, Holt RD (1997) Toward an integration of tarity. Wetlands 33:83–89 landscape and food web ecology: the dynamics of spatially subsi- Valiela I, Teal JM (1974) Nutrient limitations in salt marsh vegetation. dized food webs. Annu Rev Ecol Syst 28:289–316 In: Reimold RJ, Queen WH (eds) Ecology of Halophytes. Aca- Riera P, Richard P, Gremare A, Blanchard G (1996) Food source of demic Press, New York, pp. 547–563 intertidal nematodes in the Bay of Marennes-Oleron (France), as Vander Zanden MJ, Gratton CJ (2011) Blowin’ in the wind: reciprocal determined by dual stable isotope analysis. Mar Ecol Prog Ser airborne carbon fluxes between lakes and land. Can J Fish Aquat 142:303–309 Sci 68:170–182 Rosenblatt AE, Heithaus MR, Mather ME, Matich P, Nifong JP, Wolf PJ, Shanholtzer SF, Reimold RJ (1975) Population estimates for Ripple WJ, Silliman BR (2013) The roles of large predators in Uca pugnax (Smith, 1870) on the Duplin estuary marsh, Georgia, coastal ecosystems. Oceanography 26:156–167 USA (Decapoda Brachyura, Ocypodidae). Crustaceana 29:79–91 Schofield JA, Hagerman AE, Harold A (1998) Loss of tannins Zimmer M (2002) Nutrition in terrestrial isopods (Isopoda: Onis- and other phenolics from willow leaf litter. J Chem Ecol cidea): an evolutionary-ecological approach. Biol Rev 77:455–493 24:1409–1421 Zimmer M (2005) Cellulases. In: Graça MAS, Bärlocher F, Gess- Seiple W (1979) Distribution, habitat preferences and breeding peri- ner MO (eds) Methods to study litter decomposition: a practical ods in the crustaceans Sesarma cinereum and S. reticulatum guide. Kluwer, London, pp 249–254 (Brachyura: Decapoda: Grapsidae). Mar Biol 52:77–86 Zimmer M, Pennings SC, Buck TL, Carefoot TH (2002) Species- Seiple W, Salmon M (1987) Reproductive, growth and life-his- specific patterns of litter processing by terrestrial isopods (Isop- tory contrasts between two species of grapsid crabs, Sesarma oda: Oniscidea) in high intertidal saltmarshes and coastal forests. cinereum and S. reticulatum. Mar Biol 94:1–6 Funct Ecol 16:596–607 Silliman BR, Zieman JC (2001) Top-down control of Spartina alterni- Zimmer M, Pennings SC, Buck TL, Carefoot TH (2004) Saltmarsh flora production by periwinkle grazing in a Virginia saltmarsh. litter and detritivores: a closer look at redundancy. Estuaries Ecology 82:2830–2845 27:753–769

1 3